Logless understands the work — not the code.

1. 1

   Work activity. Commit, pull request, review and issue events from the tools you connect.

2. 2

   Minimal metadata. Titles, timestamps and types. Never file contents or diffs.

3. 3

   Logless drafts. Events are grouped into draft entries by day and project.

4. 4

   Timesheet draft. A useful starting point, never a claim of exact hours.

5. 5

   You review. Edit, add or remove anything before it counts.

6. 6

   Approved timesheet. The only thing your manager ever sees.

Logless doesn't need, and doesn't ask for: source code, diffs, screenshots, keystrokes or desktop activity.

Only lightweight metadata from the tools you explicitly connect. Today that means GitHub:

Commits

First line of the message, and timestamp.

Pull requests

Title and timestamps.

Reviews

That a review happened, on which pull request, and when.

Issues

Title and timestamps.

Logless deliberately doesn't fetch or store source code, file contents or diffs, even though GitHub's API would allow it. They aren't needed to draft a timesheet, so they aren't collected.

Logless stores the names of the repositories your team connects, because an admin has to choose which ones to track and which internal project each belongs to. That mapping screen in Settings is the only place they are shown. Your timesheet, your manager's view, CSV exports and every AI-assisted feature use your internal project names instead, never GitHub repository, organisation or branch names. Each activity record keeps an internal link to its repository so the mapping keeps working, but not the repository's URL.

When AI-assisted summaries are enabled, only an anonymised signal is sent to the AI provider: activity counts (commits, pull requests, reviews) and a time window, referenced by an opaque internal project label. Never a commit message, a pull request title, a repository name or any other identifying detail. With no AI provider configured, which is the case today, Logless works identically using fixed rules only.

Traffic to Logless is encrypted in transit (TLS/HTTPS). Your GitHub access token is encrypted at rest and never sent to the browser. This is not end-to-end encryption: Logless's servers can decrypt data to work on it, which is what lets them build your timesheet. We won't describe Logless as end-to-end encrypted, and we won't claim zero data retention from an AI provider unless the provider and plan genuinely guarantee it. Neither is true of the current setup.

This list exists because timesheet tools have a history of drifting into surveillance. Logless is built to avoid that.

• Take screenshots of your screen
• Log your keystrokes
• Track your mouse or cursor movement
• Access your webcam or microphone
• Monitor your browser history
• Monitor your desktop or IDE
• Score, rank or compare developers' productivity

You see your own detected activity and draft timesheet, and you control every entry before approving it. Managers and admins see approved timesheets, whether each person has approved yet, and aggregate hours. They don't see draft weeks, your underlying activity feed, or repository names. Company data is isolated: every query is scoped to your company, so no user can access another company's clients, projects or timesheets.

Logless runs on a small number of third-party services. Each sees only what it needs:

GitHub

The source of activity metadata, through the access each developer grants.

Vercel

Hosts the website and app, and provides aggregated page-view analytics for this site.

Neon

The Postgres database where accounts, timesheets and activity metadata are stored.

Anthropic

Only if AI summaries are switched on (they are off today). Receives counts and time windows, never titles or names.

Resend

Sends the Logless team an email when someone joins the early-access list or creates an account.

Detected activity metadata is deleted automatically after 180 days. Your approved timesheets, the hours and descriptions you reviewed and signed off, are your own records and are kept independently of the activity data that produced them.

Drafted hours are a starting estimate from fixed, deterministic rules, never a claim of exactly how long something took. Every entry is editable, and nothing is final until you approve it. Revenue recovery figures are estimates based on your own billing rates, not guarantees.

What's built today, and what we're designing towards. Nothing under Next or Later is live yet.